MAS Guidelines for Synthetic Intelligence (AI) Risk Management

Singapore, 13 November 2025… The Monetary Authority of Singapore (MAS) right this moment issued a session paper proposing a set of Guidelines on AI Risk Management (“Guidelines”) to information monetary establishments on the accountable use of AI within the monetary sector.

2 The proposed Guidelines will apply to all monetary establishments (FIs) and set out MAS’ supervisory expectations on oversight of AI threat administration in FIs, key AI threat administration techniques, insurance policies and procedures, key AI life cycle controls, in addition to capabilities and capability wanted for using AI.

3 MAS recognises that AI will be utilized to a variety of use instances, and that the dangers related to completely different utilization of AI could fluctuate based mostly on the dimensions, scope and enterprise fashions of FIs. The Guidelines goal to ascertain a set of expectations which can be usually relevant throughout the monetary sector, and could also be utilized in a proportionate method – commensurate with the scale and nature of FIs’ actions, use of AI, and their threat profiles. The Guidelines will cowl completely different AI functions and applied sciences, together with Generative AI, in addition to newer developments comparable to AI brokers.
 

4 MAS has set out expectations for FIs within the following key areas:

• Oversight of AI Risk Management. Board and senior administration of FIs play a key function within the governance and oversight of AI threat administration, together with the institution and implementation of frameworks, constructions, insurance policies and processes for AI threat administration, and fostering the suitable threat tradition for using AI.
• Key AI Risk Management Systems, Policies and Procedures. To assist oversight and threat administration of AI use, FIs want to ascertain clear identification processes for AI utilization throughout the agency, keep correct and up-to-date AI inventories, and implement threat materiality assessments that issue influence, complexity and reliance dimensions.
• AI Life Cycle Controls,
  Capabilities and Capacities. To handle the dangers of AI all through its lifecycle, FIs ought to plan for and implement strong controls in key areas comparable to information administration, equity, transparency and explainability, human oversight, third-party dangers, analysis and testing, monitoring and alter administration. Such controls must be utilized based mostly on their relevance and be proportionate to the assessed threat materiality of AI utilization. FIs also needs to make sure that their capabilities and capacities are satisfactory for his or her use of AI.

5 The Guidelines construct on MAS’ supervisory thematic overview of key banks’ use of AI in 2024, in addition to discussions with FIs.

6 Ms Ho Hern Shin, Deputy Managing Director, mentioned, “The proposed Guidelines on AI Risk Management provide financial institutions with clear supervisory expectations to support them in leveraging AI in their operations. These proportionate, risk-based guidelines enable responsible innovation by financial institutions that implement the relevant safeguards to address key AI-related risks.”

7 The public session paper is obtainable on MAS’ web site right here . MAS invitations events to submit their feedback on the proposals by 31 January 2026.

***